23. RARE CODING VARIANTS IN SCHIZOPHRENIA HAVE ADDITIVE EFFECTS ON POLYGENIC RISK AND ASSOCIATE WITH COGNITIVE ABILITIES IN INDIVIDUALS WITH PSYCHOTIC DISORDERS

The Schizophrenia Exome Sequencing Meta-Analysis (SCHEMA) consortium have identified specific genes in which ultra-rare predicted loss-of-function (pLoF) variants substantially increase schizophrenia (SZ) risk (OR > 2). However, their impact on clinical presentation and disease trajectory compared to idiopathic psychosis remains unclear. We analyzed up to 50 years of longitudinal healthcare data from 10,470 sequenced individuals with psychotic disorders in the SUPER-Finland study, focusing on carriers of pLoF variants in SZ-associated genes identified by the SCHEMA consortium. We identified 139 individuals carrying pLoF variants: 75 in exome-wide significant genes and 64 in genes likely associated with SZ (FDR < 0.05). Overall, pLoF carriers had similar diagnostic composition of psychotic disorders as non-carriers, though those with variants in the significantly associated SZ genes showed a higher SZ diagnosis rate (69.3% vs. 55.3%, p=0.027). Consistent with a liability threshold model, pLoF-carriers had lower SZ polygenic scores (SZ-PGS) than individuals with idiopathic psychosis (mean +0.60 SD vs +0.79 SD, p=0.022), but higher than population controls (p= 3.46e-12). This difference was larger in individuals with SZ diagnosis (mean +0.62 SD vs. +0.96 SD, p=6.35e-04). The association was primarily driven by the genes SETD1A, AKAP11, and XPO7, with SETD1A carriers having the lowest SZ-PGS (mean -0.16 SD, p=0.0022). Cognitive performance was markedly impaired among pLoF carriers, who made more errors on the Paired Associates Learning (PAL) test (β=0.56 [0.38–0.75], p=1.23e-09). Their performance resembled that of non-carriers with co-morbid intellectual disability (ID) (β=0.50 [0.40–0.61], n=536) and psychosis cases in SUPER who had deleterious variants in haploinsufficient ID-genes (β=0.55 [0.29-0.79], n=89). Restricting the analysis to SZ cases yielded a similar effect size for the negative impact of pLoF variants on cognition (β = 0.56 [0.31-0.81], p= 9.11e-06), primarily driven by variants in AKAP11 and XPO7. Longitudinal data also revealed that pLoF carriers experienced earlier psychiatric and non-psychiatric hospitalizations, poorer childhood well-being, lower school performance, and reduced adult social attainment (FDR < 0.05). In symptom profile analyses, carriers of pLoF variants in exome-wide significant SCHEMA genes previously linked to bipolar disorder (n=24) had affective symptom levels comparable to bipolar patients but showed a significantly higher psychotic symptom burden (p < 0.001). In contrast, carriers of pLoF variants in genes only known to associate with SZ showed no increase in affective symptoms and had the lowest ratio of affective to psychotic symptoms across all groups, including non-carriers with SZ (p < 0.001). Carriers of rare pLoF variants in SZ risk genes show markedly worse cognitive functioning and earlier impairments across multiple domains of life. Their lower polygenic burden for SZ supports an additive liability model, in which rare, high-impact variants can compensate for lower common variant burden. Importantly, our findings suggest that pLoF variants in specific genes are associated with distinct symptom trajectories within the psychosis spectrum. These gene-specific differences underscore the need to investigate underlying biological mechanisms that contribute to symptom dimensions, rather than categorical diagnoses alone.